257 



occur when a word whose stem contains a voiced fricative is 

 either inflected or supplied with a suffix so that it undergoes 

 a shifting of stress, as for example : 



iioik a blade of grass plur. ^ifit 



al^oq a sole of a foot or a boot » aX'ut 

 ni^^aq (SGr.) a snare » nHy'àt 



neriwoq is eating n^ep'iw^ik the place where you 



eat, table or dining-room 



Now if we find similar relations existing between the same 

 kinds of sounds as they occur in two different dialects, it is 

 natural to regard them after the analogy of the parallel rela- 

 tions within the single dialect. So v/hen we find that kiliqwaq 

 fossil elephant, mammoth (Ray, Petitot) in the western dialects has 

 become kilifaq in Greenlandic, the latter form is undoubtedly 

 later than the former; the unvoiced fricative f in Greenlandic 

 has accordingly originated by the assimilation of two different 

 voiced fricatives, klwqaq a servant, a house-maid in the Labrador 

 and Mackenzie dialects has got the form ki"'<p-aq in SGr., 

 where the voice perhaps still lingers a little in the beginning 

 of the y7-sound, but otherwise the case is essentially the same 

 as the preceding one, only that w in the unassimilated form 

 here stands before q. In the SWAl. dialect I find the word 

 in the form kü'vügäk messenger, envoy (Barnum). — In the pre- 

 ceding sections (pp. 225, 233, 236, 246, 248), there are examples 

 showing the same relations between the other open consonants. 

 The orthography in the specimens of the western dialects has 

 throughout, as we have seen, bw, ivq^ Iq, ^;, rj, rg, as against 

 Greenlandic y?, Å, ç, y. Therefore I have been led to infer 

 that voiced sounds originally corresponded to the present un- 

 voiced fricatives in SW. Greenlandic. 



In these cases, Upernawik Greenlandic and Ammassalik 

 Greenlandic, as has often before been mentioned, have got 

 unvoiced stopped consonants, for example: ki^pa-t instead of 

 x.xxi. 17 



